Biodegradable micro-osmotic pump for long-term and controlled release of basic fibroblast growth factor

NANOTECHNOLOGY

Available online at www.sciencedirect.com

Journal of Controlled Release 124 (2007) 98 – 105 www.elsevier.com/locate/jconrel

Biodegradable micro-osmotic pump for long-term and controlled release of basic fibroblast growth factor WonHyoung Ryu a,⁎, Zhinong Huang b , Fritz B. Prinz a , Stuart B. Goodman b , Rainer Fasching a a

Rapid Prototyping Laboratory, Mechanical Engineering Department, 440 Escondido Mall, Bldg. 530, Rm 226, Stanford University, Stanford, CA 94305, USA b Department of Orthopaedic Surgery, School of Medicine, R153, 200 Pasteur Dr, Stanford University, Stanford, CA 94305, USA Received 26 June 2007; accepted 17 August 2007 Available online 25 August 2007

Abstract Microelectromechanical system (MEMS) technology not only provides the possibility of integration of multiple functions but also enables more precise control of dosing of therapeutic agents when the therapeutic window is very limited. Local delivery of basic fibroblast growth factor (bFGF) over a specific dose and time course is critical for mesenchymal tissue regeneration. However, bFGF is degraded quickly in vivo and difficulty of controlling the dose level impedes its effective use in angiogenesis and tissue regeneration. We constructed biodegradable microosmotic pumps based on MEMS technology for long-term controlled release of bFGF. The devices were constructed by micro-molding and thermal assembly of 85/15 poly(L-lactide-co-glycolide) sheets. The release of bFGF was regulated at 40 ng/day for four weeks; bioactivity was assessed by monitoring the growth of 3T3 fibroblasts. The proposed devices can be further miniaturized and used for the delivery of multiple therapeutic agents at the individual releasing schedules. © 2007 Elsevier B.V. All rights reserved. Keyword: Controlled drug delivery; MEMS; Micro-fabrication; Osmotic pump; Biodegradable polymers; PLGA; Basic fibroblast growth factor

1. Introduction Micro- and nanosystems for drug delivery offer the possibility of integration with diagnostic functions or implants and allow precise dosage control for highly potent drugs [1–3]. Recent examples include micro-particles for oral delivery [4,5], micro-needles for transdermal delivery [6], micro-fluidic probes for convection enhanced drug delivery [7], micro-pumps based on osmosis [8] or piezoelectricity [9], electro-chemically controlled micro-chips [10,11], and micro-fabricated nanochannel systems [12]. Even though micro-fabricated materials for these microelectromechanical system (MEMS) drug delivery devices showed biocompatibility and reduced biofouling [13], the nondegradability of these devices is still disadvantageous in comparison to devices made of biodegradable materials. There have been great advances in delivery devices made of biodegradable polymers. However, most of these relied on polymer erosion for controlled delivery [14–20]. As this ⁎ Corresponding author. Tel.: +1 650 723 0084; fax: +1 650 723 5034. E-mail address: [email protected] (W. Ryu). 0168-3659/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2007.08.024

erosion-based control depends on the property of polymers, it limits effective treatment of a disease when more delicate control of drug release is necessary. MEMS technology has shown great promise for more versatile and precise control of many chemicals. Nonetheless, less work of MEMS has been done on biodegradable materials, mainly due to the absence of appropriate technologies. A few examples include sandwich devices made of a biodegradable polymer by 3D printing [21] and micro-reservoir devices with the biodegradable membranes for pulsatile delivery [22]. However, there still remains great need for further improvement of MEMS-based drug delivery devices. In particular, more attention needs to be paid to the delivery of relatively unstable therapeutic molecules such as peptides and growth factors. In the United States, approximately 800,000 surgical procedures are performed annually in which some form of bone graft is used [23]. Autograft bone has been used in diverse clinical situations such as in spinal fusion, fractures, non-unions, revision total joint replacements, and tumors. However, autograft bone is limited in supply and the morbidity associated with harvesting of bone from a remote site may be substantial [24–

29]. Bone graft substitutes and extenders attempt to limit the use of autologous bone grafts. The use of allograft bone, ceramics, demineralized bone matrix, and composite grafts may become more effective when it is combined with bone derived growth factors [30–32]. Basic fibroblast growth factor (bFGF) is a protein with autocrine and paracrine effects, and affects numerous cell types in different systems in the body. In bone, bFGF is produced by cells of the osteoblastic lineage. Its major effects on bone include the induction of neovascularization, and modulation of osteoblastic proliferation and differentiation. It has been shown that addition of bFGF enhances the incorporation of allograft bone in rats in a dose- and time-dependent manner and accelerates the healing of fractures in rabbits [33,34]. For these reasons, local delivery of bFGF could potentially be advantageous in clinical scenarios of bone loss and repair, such as fracture healing, periprosthetic osteolysis, etc. Ideally, delivery devices of bFGF can be placed proximate to fracture sites by wrapping the devices around the fractured bones during the surgery, or incorporating the devices into implants during primary or revision total hip replacement for enhancing oseeointegration and reducing osteolysis. Due to the relatively short halflife time of bFGF in vivo [35], effort has been focused on the delivery of bFGF over a more prolonged time period [36–42]. However, most of the methods have delivered bFGF over only a few days or required complex processes to synthesize delivery vehicles. We have developed biodegradable MEMS devices for delivery of growth factors for a prolonged time period. The MEMS devices are comprised of a drug reservoir and microchannels which connect between the reservoir and the open end of the micro-channels. The MEMS device operates by either osmotic or diffusion pumping depending on the choice of geometric parameters and water permeability of polymers used. In this study, stronger osmosis was induced by enlarging the area of the water-permeable membrane and incorporating highly potent osmotic agents (polyethylene glycols). The diffusional delivery was suppressed by elongating the micro-channels. The release of the bFGF from the MEMS devices was maintained at a constant level of 40 ng/day on average over four weeks. The corresponding bioactivity of the released bFGF was maintained 80% except during the first week (30% bioactivity). In addition, the effect of micro-channel length on the release profiles of the devices was investigated. It was shown that elongation of micro-channels down-regulated the release of fluorescent red dye for two weeks.

99

Fig. 1. Device configuration and principles of operation. The device has a microfabricated reservoir and four micro-channels in a biodegradable polymer layer. This layer is sealed by another biodegradable polymer layer. Drugs and osmotic agents are loaded in the reservoir. Water flows into the reservoir by osmosis and a hydrostatic pressure builds up in the reservoir leading to the inflation of the device. This pressure drives drug molecules through release channels to the outside of the device.

for drugs and micro-channels (50 × 50 μm2) for release of the drugs. Another layer (25 μm thick) of the 85/15PLGA acts as a semi-permeable layer (water permeability, k = 1 × 10− 21 m2/ Pa s). Once the device is immersed in an aqueous environment, osmotic pressure drives water into the reservoir area, as shown in Fig. 1. This inflates the reservoir and the internally-built pressure begins to propel drugs through the microchannels. The release rate from the proposed devices can be estimated from an osmotic pumping principle [43]. When the device operates in the steady state condition, the osmotic delivery rate of any substance contained in the devices is given by:     dm dV A ¼ d C ¼ kd Dpd C dt osmotic dt osmotic h

ð1Þ

where dm/dt is the osmotic release rate; dV/dt is the volume flux of a solvent; Δπ is the osmotic pressure; C is the concentration of a substance to be delivered; k is the water permeability; A is the membrane area; h is the membrane thickness. Once the device is fully inflated by the in-flux of water through the reservoir membrane, the solution in the reservoir starts flowing through the micro-channels. This solution flow through the micro-channels is governed by Poiseuille's law [44]:

2. Materials and methods 2.1. Device design and principle

dV wh3 DP ¼ dt 12l L

A thin biodegradable device as an osmotic pump was designed for sustained and constant release of growth factors. The pumping device was micro-fabricated on a layer of a synthetic biodegradable polymer by micro-molding. A diagram of the device is shown in Fig. 1. The bottom layer (75 μm thick) contains a reservoir (A = 2 cm2, depth = 50 μm)

where dV/dt is the volume flow rate; w, h, and L are the width, depth, and length of a channel, respectively; μ is the viscosity of a solution; ΔP is the pressure difference between the inside and outside of the device. As long as the ΔP remains constant, the release profile can be modulated by the change of the geometric parameters.

ð2Þ

NANOTECHNOLOGY

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

NANOTECHNOLOGY

100

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

2.2. Fibroblast growth factor and red dye Recombinant mouse bFGF (PMG0031, Biosource, Camarillo, CA) was formulated in a phosphate-buffered saline (PBS) solution containing bovine serum albumin (0.1% w/v, Sigma, St Louis, MO) and heparin (600 μg/mL, Sigma, St Louis, MO). The final concentration of bFGF in the stock solution was 600 μg/mL. Polyethylene glycols of two different molecular weights (PEG600, Mw = 600, and PEG6000, Mw = 6000, Lancaster Synthesis, Pelham, NH) were purchased. The stock solution of bFGF was mixed in a mixture of the PEGs (CPEG600 = CPEG6000 = 0.4 g/mL). The final mixture had the bFGF concentration of 85 μg/mL. Fluorescent red dye (105403 Fluorescent FWT Red, Bright Dyes, Miamisburgh, Ohio) was formulated together with the PEG mixture at a concentration of 0.34 mg/mL. 2.3. Device fabrication Micro-molds were fabricated by a combination of wet and reactive ion etching (RIE) processes. The detailed process of the mold fabrication was reported previously [45,46]. 85/15poly( L -lactide-co-glycolide) (85/15PLGA, IV = 2.3) was obtained from Purac. The polymer was compression molded into 25 and 75 μm thick films using a Carver Auto C press (Carver Inc, Indiana). The reservoir and micro-channels were micro-molded on 75 μm thick films at 160 °C under 1.9 MPa for 2 min in the Carver press (Fig. 2). The detailed process of micro-molding was described previously [45,46]. The micro-structured layer and the other plain layer (25 μm thick film) were stacked together with a thin Teflon® sheet (McMaster-Carr, Los Angeles, CA) inserted in the reservoir area between the two layers. This assembly was thermally bonded at 80 °C under 345 kPa for 5 min. Later, the Teflon® sheet was removed from the assembled device, leaving the reservoir area open for loading of fibroblast growth factor and a fluorescent dye (Fig. 2). A mixture of either the PEG/fibroblast growth factor or the PEG/fluorescent red dye was loaded into the reservoir by a syringe (20 μL per device). Then, the device was sealed by a heat sealer (Model AN90, Andersen Products, Haw River, NC) at the edge of the device. The assembled devices were rolled to fit into the wells of micro-plates for incubation. 2.4. In vitro release measurement The amount of bFGF in the released medium was determined by an ELISA kit (Quantikine human FGF basic Immunoassay, R&D Systems, Minneapolis, MN). The optical densities were measured using a micro-plate reader (Spectra MAX M2, Molecular Devices) at 450 nm and 595 nm. Biological activity of bFGF from devices over four weeks was assessed by measuring the ability to stimulate cell proliferation. CellTiter 96® AQueous Non-Radioactive Cell Proliferation assay kit (Promega Corp., Madison, WI) was used to determine the bioactivity of bFGF. The manufacturer's protocol was followed carefully. Briefly, samples and bFGF standards were incubated

Fig. 2. Process of device fabrication.

with 5000 BALB/c 3T3 clone A31 cells (CCL 163, ATCC, Manassas, VA) in 50 μL F12/DME media (Invitrogen, supplemented with 10 μg/mL insulin, 10 μg/mL transferrin, 10 ng/mL selenium, 100 μg/mL ovalbumin, 1 μM dexamethasone and 5 μg/mL fibronectin) at 37 °C in a 5% CO2, humidified incubator. After 44 h, 20 μL of fresh MTS/PMS solution was added into the media and incubated at 37 °C for 2 h. The absorbance of the media was recorded at 490 nm. The bioactivity was converted to the equivalent amount of freshreconstituted bFGF using the standard curve generated at the same time. The amount of red dye released from the devices was measured with excitation at 550 nm and emission reading at 594 nm.

Fig. 3. Picture of a sample device and the zoomed-in view of a micro-channel. Channels are 50 μm wide and 50 μm deep.

2.5. Statistical analysis Statistical analysis was performed using an ANOVA in Microsoft Excel® 2002 (Microsoft Corp., Redmond, WA). A value of p b 0.05 was considered statistically significant. 3. Results 3.1. Device fabrication Drug reservoirs and micro-channels were precisely formed on the layers of 85/15PLGA by the micro-molding process (Fig. 3). An assembled device had a reservoir filled with a mixture of bFGF and PEG, and four distinct micro-channels for the release of bFGF (Fig. 3). The micro-channels were wellpreserved without deformation during thermal assembly. Sealing around the reservoirs and micro-channels was seamless such that no leakage or cleavage was observed during the experiments with color dyes. 3.2. Prediction of delivery rate

101

of micro-channels. The diffusion coefficient of lactalbumin (D =13 × 10− 7 cm2/s) was used for the value of DbFGF in the calculation based on the similarity to the molecular weight of bFGF [48]. When a device has the release channels of L = 2 mm and Ac = 10− 8 m2, the diffusion-based release rate of bFGF, (dm/dt)diffusion, was calculated as 2.5 ng/day. Since this is only about 3% of the osmotic release, the diffusional release of bFGF from the proposed device can be ignored. The osmotic release rate of fluorescent red dye was also estimated from Eq. (1) with the following parameters: Cdye = 170 μg/mL, C PEG600 = 0.2 g/mL, C PEG6000 = 0.2 g/mL, Δπ = 3 MPa. The calculated volume flow rate (dV/dt) was 1.9 μL/day and the corresponding release rate of the red dye (dm/dt) was 323 ng/day. 3.3. Stability of bFGF in release medium The bFGF released from devices stayed in its release medium for a maximum of either two or four days depending on the sampling frequency. When bFGF stays in the release medium for a longer period time, the bFGF stability can decrease due to the low concentration environment of the release medium. Thus, in order to verify the effect of the sampling frequency on bFGF stability, the ELISA response of bFGF in the release medium over time was monitored. The solutions of bFGF sampled at day 6 (bFGF stayed in the release medium for a maximum of two days) were further incubated for an additional two days at 37 °C. The ELISA response of the samples showed that the additional two days in the release medium reduced the ELISA response by 57 ± 10% (n = 4). This indicates that the sampling frequency of the bFGF should be taken into consideration when the released bFGF is quantified by ELISA.

The osmotic release rate of bFGF was estimated from Eq. (1). Each device had 20 μL of the mixture of bFGF solution (85 μg/mL), PEG600 (0.4 g/mL), and PEG6000 (0.4 g/mL). The mixture in the device was diluted double when the water uptake by osmosis was stabilized. The dilution being considered, the concentration of each substance in the device can be described by the following parameters: CbFGF = 42.5 μg/mL, CPEG600 = 0.2 g/mL, and CPEG6000 = 0.2 g/mL. The osmotic pressure (Δπ) was estimated to be about 2 MPa based on an empirical formula reported elsewhere [47]. The water permeability (k) of the 85/15PLGA was measured as 1 × 10− 21 m2/Pa s. By using Eq. (1), the volume flow rate (dV/dt) was calculated as 1.9 μL/day and the release rate (dm/dt) of bFGF was calculated as 80 ng/day. The release rate of bFGF by diffusion mechanism was also estimated by applying the following:   dm CbFGF d Ac ¼ DbFGF d dt diffusion L

ð3Þ

where DbFGF is the diffusion coefficient of bFGF; L is the length of micro-channels; Ac is the cross-sectional area

Fig. 4. Release profile of bFGF from sample devices. Left axis represents daily release amount in ng/day (♦) and right axis represents cumulative release amount in percentage (■). The decrease in the release rate after day 8 is due to the degradation of bFGF in the release medium from the change of the sampling interval from two to four days. Estimated release rate by Eq. (1) was 80 ng/day (---).Values represent mean ± standard error (n = 4). ⁎ p b 0.05 compared to release samples from day 12 to day 28 (Student t-test).

NANOTECHNOLOGY

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

NANOTECHNOLOGY

102

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

3.4. Release kinetics of bFGF and bioactivity The release of bFGF over four weeks was measured using ELISA (Fig. 4). No initial burst was observed. By day 8, the release of bFGF was controlled at the rate of 40 ng/day and was maintained at a constant level of 20 ng/day for the next three weeks. When the degradation of bFGF in a low concentration release medium is considered, the decreased release rate after day 8 is considered due to the change of the sampling period from two to four days. Thus, the actual amount of bFGF released from the devices was maintained constant at 40 ng/day over the whole four-week period. The difference between the predicted (80 ng/day) and actual release rate (40 ng/day for week 1, 20 ng/day for weeks 2–4) is explained by degradation of bFGF in the release medium after release from the devices. During the four-week period, a total of 42% of the loaded bFGF was quantified by ELISA. After this four-week experiment, some bFGF remains in the devices since the release rate was maintained at a constant level rather than gradually decreasing. The biological activity of the growth factors released from the devices was assessed by measuring the ability to stimulate the proliferation of fibroblasts (BALB/c 3T3 clone A31) in the released medium collected. The proliferation of fibroblasts was quantified by optical density of each sample medium. The measured values were converted to the equivalent amounts of freshly reconstituted bFGF by using a standard curve generated at the same time. The equivalent values as indicators of bioactivity were plotted over four weeks (Fig. 5). This plot represents the equivalent delivery of freshly reconstituted bFGF when it was manually dosed at each time point. The biological activity of the released medium was equivalent to a daily injection of 15 ng of fresh-reconstituted bFGF (15 ng/day) on average. The variation among samples at different time points was not significant (p N 0.05). 3.5. Effect of micro-channel length on release profiles The osmotic delivery of drugs is regulated mainly by the water permeability, size, and thickness of the membrane of the

Fig. 5. Equivalent Bioactivity of bFGF released from sample devices. Values represent mean ± standard error (n = 4). There was no significant difference (p N 0.05) between samples at each time point.

Fig. 6. Effect of micro-channel length on release rate. Release rate of drug slows down as micro-channels become longer due to the increased hydraulic resistance. ♦: 2 mm, ■: 4 mm, ▲: 8 mm long channels. —: estimated release rate by Eq. (1) for 2 mm long channels.

devices [43]. The osmotic release rate in Eq. (1) is obtained only when the volume of water in-flux equals to the volume of the solution released from a device. Typically, the alteration of the exit orifice condition does not affect the release rate of osmotic devices, unless the release is governed by diffusion [43]. However, under osmotic delivery, the modulation of release rate by the exit orifice condition is possible when there exists either relatively large hydrostatic pressure to decrease the osmotic influx of water or pressure leak through the membrane of devices which leads to a constant pressure within the reservoir. In order to understand the effect of the exit orifice condition, the lengths of micro-channels were varied (2, 4, and 8 mm). Fluorescent red dye was used as a model drug in this experiment. The release profiles of the dye for micro-channels with three different lengths show dependency on the lengths of the channels for two weeks (Fig. 6). The device with the shortest channel released the dye at a rate almost comparable to the estimated value (323 ng/day) from the Eq. (1). The release rate decreased when the channels became longer. The elongation of the channels increases the hydrostatic resistance and slows down the transport of molecules (Eq. (2)). This can also be understood by the increased time constant (τ = RC, R is flow resistance of channels and C is volumetric capacitance of a reservoir) from the perspective of an equivalent circuit model. This increased time constant delays the equilibration of the release rate of a device with longer channels. However, the release rates from devices with different lengths of channels also showed three different plateaus of the release rate after one week, indicating the change of the length affected the steady state condition of the devices. One possible interpretation is that the presence of air trapped in the reservoir mitigated further build-up of pressure in the devices by permeating through the membrane of the reservoirs. This allowed a constant pressure difference between the inside and outside of the devices and longer channels could induce higher flow resistance and decrease the overall release rate in the steady state (Eq. (2)).

4. Discussion The release of bFGF from the biodegradable osmotic pumps was regulated at the desired level for a month-long period. The device has two modes of delivery: osmotic and diffusional delivery. While both modes contribute to the delivery of drugs, the portion of each mode can be adjusted by changing the micro-geometries of the devices. For example, the diffusional delivery depends on the cross-sectional area (Ac) and the lengths (L) of micro-channels (Eq. (3)). Making the delivery channels smaller and longer minimizes the delivery by diffusion of drugs and maximizes osmotic contribution to the total delivery rate. This effect of micro-geometry on delivery modes can be represented by a ratio of osmotic over diffusional delivery. The ratio is expressed by the following: m osmotic ¼ m diffusion



Am Ac



kdDp D

  L i0:016L h

ð4Þ

where L is expressed in the unit of [μm] and other parameters were fixed as used in this experiment. The change of the operation mode for a given length (L) of release channels is illustrated (Fig. 7). For the length of 60 μm, the contribution from each mechanism is balanced (balance point). As the ratio becomes larger than one, the osmotic mechanism is more dominant. Likewise, diffusion becomes more dominant when the ratio becomes smaller than one. In this experiment, channel length of 2 mm (L = 2 mm, operation point) was used to minimize the diffusional delivery (ratio = 32). This resulted in the estimated osmotic delivery of 80 ng/day and diffusional delivery of 2.5 ng/day. The stabilization of bFGF in a device was achieved by keeping bFGF in a dense mixture of PEG600/PEG6000 in the reservoir of a device. As long as bFGF stayed in a reservoir full of a PEG mixture, the degradation of bFGF was minimized. Degradation of bFGF occurred only when bFGF was released from the device chamber to the dilute release medium. The released bFGF in the buffer medium lost about half of the

103

sensitivity to ELISA over two days. Thus, the sudden drop of the release rate from 40 to 20 ng/day after day 8 is due to the change of the sampling period from every two to four days. This indicates the use of PEG mixture for bFGF in a drug reservoir effectively prolonged the stability of bFGF for the entire experiment period. The quantitative measurement of bFGF stability in the chamber of a device is currently under further investigation in our laboratory. Control of the release profile by micro-geometry of drug delivery devices is very attractive when more complex dosing schedules of multiple therapeutic agents is required. The release profile can be pre-programmed to the device by designing a proper micro-geometry for one type of dose schedule without a need for developing a new formulation for the specific schedule. For example, when a larger dose rate than 40 ng/day is needed for a study of animals larger than mice or rats, the increased rate can be achieved either by enlarging the reservoir membrane of an osmotic-driven device or by increasing the cross-sectional area of the channels of a diffusional-driven device. Although only a single drug was released from a single device in this experiment, more complex delivery schemes can also be facilitated by this technology. For example, a single device can contain two drug reservoirs with individual release channels. Having two different sizes of channels (Ac or L) or reservoir membranes (A) will allow two different release rates from the two reservoirs. Ultimately, this enables a combinatorial therapy with multiple drugs which require multiple dosing schedules and rates from a single device. Unlike conventional delivery systems made of biodegradable polymers, the release mechanism of the biodegradable osmotic pumps was de-coupled from the degradation of polymers. The release was controlled only by the micro-geometries shaped in the devices and the permeability of a polymer. Degradation of devices occurred after the bFGF release was completed. This de-coupling of the release from degradation of devices offers great advantages over degradation-based delivery systems. The advantages include more accurate and easier modulation of release, avoidance of drug exposure to an acidic environment from degradation [49,50], reduction of adverse inflammatory response during the release [51,52], and minimization of any possible reaction between drugs and the degraded polymers [53–55]. 5. Conclusions A novel concept of a planar shaped biodegradable osmotic pump has been developed for long-term and controlled delivery of large therapeutics molecules such as peptides and growth factors. The osmotic pumping devices were constructed by micro-fabrication of 85/15PLGA layers. The functionality of the proposed concept was shown by delivering bFGF at the constant rate of 40 ng/day for a four-week period in vitro.

Fig. 7. Osmotic vs. diffusional delivery. Principal mode of delivery can switch between osmotic and diffusion modes by adjusting the length of micro-channels. Sample devices operated under osmotic mode with L = 2 mm (operation point). When L = 60 μm (balance point), diffusion and osmotic modes contribute equally to the delivery of drug.

Acknowledgements Funding is gratefully acknowledged from the Musculoskeletal Transplant Foundation.

NANOTECHNOLOGY

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

NANOTECHNOLOGY

104

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105

References [1] D.A. LaVan, T. McGuire, R. Langer, Small-scale systems for in vivo drug delivery, Nat. Biotechnol. 21 (10) (2003) 1184–1191. [2] S.Z. Razzacki, P.K. Thwar, M. Yang, V.M. Ugaz, M.A. Burns, Integrated microsystems for controlled drug delivery, Adv. Drug Deli Rev. 56 (2004) 185–198. [3] Y. Lu, S.C. Chen, Micro and nano-fabrication of biodegradable polymers for drug delivery, Adv. Drug Deli Rev. 56 (2004) 1621–1633. [4] S.L. Tao, M.W. Lubeley, T.A. Desai, Bioadhesive poy(methyl methacrylate) microdevices for controlled drug delivery, J. Control. Release 88 (2003) 215–228. [5] S.L. Tao, T.A. Desai, Micromachined devices: the impact of controlled geometry from cell-targeting to bioavailability, J. Control. Release 109 (2005) 127–138. [6] J. Park, M.G. Allen, M.R. Prausnitz, Biodegradable polymer microneedles: fabrication, mechanics and transdermal drug delivery, J. Control. Release 104 (2005) 51–66. [7] K.B. Neeves, C.T. Lo, C.P. Foley, W.M. Saltzman, W.L. Olbricht, Fabrication and characterization of microfluidic probes for convection enhanced drug delivery, J. Control. Release 111 (2006) 252–262. [8] Y. Su, L. Lin, A water-powered micro drug delivery system, J. Microelectromechanical Syst. 13 (1) (2004) 75–82. [9] D. Mailefer, H. van Lintel, G. Rey-Mermet, R. Hirschi, A highperformance silicon micropump for an implantable drug delivery system, Proc. IEEE MEMS 1999 (1999) 541–546. [10] J.T. Santini, M.J. Cima, R. Langer, A controlled-relesae microchip, Nature 397 (1999) 335–338. [11] Y. Li, H.L.H. Duc, B. Tyler, T. Williams, M. Tupper, R. Langer, H. Brem, M.J. Cima, In vivo delivery of BCNU from a MEMS device to a tumor model, J. Control. Release 106 (2005) 138–145. [12] F. Martin, R. Walczak, A. Boiarski, M. Cohen, T. West, C. Cosentino, M. Ferrari, Tailoring width of microfabricated nanochannels to solute size can be used to control diffusion kinetics, J. Control. Release 102 (2005) 123–133. [13] G. Viskerician, M.S. Shive, R.S. Shawgo, H. von Recum, J.M. Anderson, M.J. Cima, R. Langer, Biocompatibility and biofouling of MEMS drug delivery devices, Biomaterials 24 (2003) 1959–1967. [14] S.S. Shah, Y. Cha, C.G. Pitt, Poly(glycolide acid-co-DL-lactic acid): diffusion or degradation controlled drug delivery ? J. Control. Release 18 (1992) 261–270. [15] R. Langer, Polymer-controlled drug delivery systems, Acc. Chem. Res. 26 (1993) 537–542. [16] P. Chen, Y.J. Park, L. Chang, D.S. Kohane, R.H. Bartlett, R. Langer, V.C. Yang, Injectable microparticle-gel system for prolonged and localized lidocaine release. I. In vitro characterization, J. Biomed. Mater. Res. 70A (2004) 412–419. [17] R.K. Agarwal, D.H. Robinson, G.I. Maze, R.A. Reinhardt, Development and characterization of tetracycline-poly(lactide/glycolide) films for the treatment of periodontitis, J. Control. Release 23 (1993) 137–146. [18] N. Faisant, J. Siepmann, J.P. Benoit, PLGA-based microparticles: elucidation of mechanisms and a new, simple mathematical model quantifying drug release, Eur. J. Pharm. Sci. 15 (2002) 355–366. [19] A. Charlier, B. Leclerc, G. Couarraze, Release of mifepristone from biodegradable matrices: experimental and theoretical evaluations, Int. J. Pharm. 200 (2000) 115–120. [20] A. Frank, S.K. Rath, S.S. Venkatraman, Controlled release from bioerodible polymers: effect of drug type and polymer composition, J. Control. Release 102 (2005) 333–344. [21] B.M. Wu, S.W. Borland, R.A. Giordano, L.G. Cima, E.M. Sachs, M.J. Cima, Solid free-form fabrication of drug delivery devices, J. Control. Release 40 (1996) 77–87. [22] A.C.R. Grayson, I.S. Choi, B.M. Tyler, P.P. Wang, H. Brem, M.J. Cima, R. Langer, Multi-pulse drug delivery from a resorbable polymeric microchip device, Nat. Mater. 2 (2003) 767–772. [23] W.W. Tomford, Overview, in: C.T. Laurencin (Ed.), Bone Graft Substitutes, ASTM, West Conshohocken, PA, 2003. [24] E.M. Younger, M.W. Chapman, Morbidity at bone graft donor sites, J. Orthop. Trauma 3 (1989) 192–195.

[25] B.L. Fowler, B.E. Dall, D.E. Rowe, Complications associated with harvesting autogenous iliac bone graft, Am. J. Orthop. 24 (1995) 895–903. [26] E.D. Arrington, W.J. Smith, H.G. Chambers, A.L. Bucknell, N.A. Davino, Complications of iliac crest bone graft harvesting, Clin. Orthop. 329 (1996) 300–309. [27] J.A. Goulet, L.E. Senunas, G.L. DeSilva, M.L. Greenfield, Autogenous iliac crest bone graft. Complications and functional assessment, Clin. Orthop. 339 (1997) 76–81. [28] N.M. Hill, J.G. Horne, P.A. Devane, Donor site morbidity in the iliac crest bone graft, Aust. N. Z. J. Surg. 69 (1999) 726–728. [29] N.A. Ebraheim, H. Elgafy, R. Xu, Bone-graft harvesting from iliac and fibular donor sites: techniques and complications, J. Am. Acad. Orthop. Surg. 9 (2001) 210–218. [30] J.S. Wang, P. Aspenberg, Basic fibroblast growth factor infused at different times during bone graft incorporation. Titanium chamber study in rats, Acta Orthop. Scand. 67 (1996) 229–236. [31] J.S. Wang, P. Aspenberg, Basic fibroblast growth factor increases allograft incorporation. Bone chamber study in rats, Acta Orthop. Scand. 65 (1994) 27–31. [32] M.L. Radomsky, T.B. Aufdemorte, L.D. Swain, W.C. Fox, R.C. Spiro, J.W. Poser, Novel formulation of fibroblast growth factor-2 in a hyaluronan gel accelerates fracture healing in nonhuman primates, J. Orthop. Res. 17 (1999) 607–614. [33] J.S. Wang, P. Asenberg, Basic fibroblast growth factor enhances bone-graft incorporation: dose and time dependence in rats, J. Orthop. Res. 14 (1996) 316–323. [34] M.L. Radomsky, A.Y. Thompson, J.W. Poser, Potential role of fibroblast growth factor in enhancement of fracture healing, Clin. Orthop. Relat. Res. 355S (1998) 283–293. [35] G.F. Whalen, Y. Shing, J. Folkman, The fate of intravenously administered bFGF and the effect of heparin, Growth Factors 1 (1989) 157–164. [36] Y. Tabata, S. Hijikata, Y. Ikada, Enhanced vascularization and tissue granulation by basic fibroblast growth factor impregnated in gelatin hydrogels, J. Control. Release 31 (1994) 189–199. [37] M. Ishihara, K. Obara, T. Ishizuka, M. Fujita, M. Sato, K. Masuoka, et al., Controlled release of fibroblast growth factors and heparin from photocrosslinked chitosan hydrogels and subsequent effect on in vivo vascularization, J. Biomed. Mater. Res. 64A (2003) 551–559. [38] O. Jeon, S.H. Ryu, J.H. Chung, B.S. Kim, Control of basic fibroblast growth factor release from fibrin gel with heparin and concentrations of fibrinogen and thrombin, J. Control. Release 105 (2005) 249–259. [39] O. Jeon, S.W. Kang, H.W. Lim, J.H. Chung, B.S. Kim, Long-term and zero-order release of basic fibroblast growth factor from heparinconjugated poly(L-lactide-co-glycolide) nanospheres and fibrin gel, Biomaterials 27 (2006) 1598–1607. [40] A. Kanematsu, A. Marui, S. Yamamoto, M. Ozeki, Y. Hirano, M. Yamamoto, O. Ogawa, M. Komeda, Y. Tabada, Type I collagen can function as a reservoir of basic fibroblast growth factor, J. Control. Release 99 (2004) 281–292. [41] J.S. Lee, D.H. Go, J.W. Bae, S.J. Lee, K.D. Park, Heparin conjugated polymeric micelle for long-term delivery of basic fibroblast growth factor, J. Control. Release 117 (2007) 204–209. [42] J. Guan, J.J. Stankus, W.R. Wagner, Biodegradable elastomeric scaffolds with basic fibroblast growth factor release, J. Control. Release 120 (2007) 70–78. [43] F. Theewues, Elementary osmotic pump, J. Pharm. Sci. 64 (1975) 1987–1991. [44] F.M. White, Viscous Fluid Flow, McGraw-Hill, New York, 1991. [45] W. Ryu, R. Fasching, M. Vyakarnam, R.S. Greco, F.B. Prinz, Microfabrication technology of biodegradable polymers for interconnecting micro-structures, J. Microelectromechanical Syst. 15 (2006) 1457–1465. [46] W. Ryu, S. Min, K. Hammerick, M. Vyakarnam, R.S. Greco, F.B. Prinz, R. Fasching, Micro-fabricated tissue scaffolds of biodegradable polymers, Biomaterials 28 (2007) 1174–1184. [47] R.P. Rand. (2007) http://www.brocku.ca/researchers/peter_rand/osmotic/ osfile.html. [48] W.M. Saltzman, Drug Delivery:Engineering Principles for Drug Therapy, Oxford University Press, New York, NY, 2001.

[49] K.A. Aathanasiou, C.M. Agrawal, F.A. Barber, S.S. Burkart, Orthopaedic applications for PLA–PGA biodegradable polymers, Arthrosc: J. Arthrosc. Relat. Surg. 14 (1998) 726–737. [50] J. Suganuma, H. Alexander, Biological response of intramedullary bone to poly(L-lactic acid), J. Appl. Biomater. 4 (1993) 13–27. [51] A.U. Daniels, M.S. Taylor, K.P. Andriano, J. Heller, Toxicity of absorbable polymers proposed for fracture fixation devices, Trans. Orthop. Res. Soc. 17 (1992) 88. [52] M.S. Taylor, A.U. Daniels, K.P. Andriano, J. Heller, Six bioabsorbable polymers: in vitro acute toxicity of accumulated degradation products, J. Appl. Biomater. 5 (1994) 151–157.

105

[53] W. Lu, T.G. Park, Protein release from poly(lactic-co-glycolic acid) microspheres: protein stability problems, PDA J. Pharm. Sci. Technol. 49 (1995) 13–19. [54] T.G. Park, W. Lu, G. Crotts, Importance of in vitro experimental conditions on protein release kinetics, stability and polymer degradation in protein encapsulated poly( D , L -lactic acid-co-glycolic acid) microspheres, J. Control. Release 33 (1995) 211–222. [55] M. Igartua, R.M. Hernandez, A. Esquisabel, A.R. Gascon, M.B. Calvo, J.L. Pedraz, Stability of BSA encapsulated into PLGA microspheres using PAGE and capillary electrophoresis, Int. J. Pharm. 169 (1998) 45–54.

NANOTECHNOLOGY

W. Ryu et al. / Journal of Controlled Release 124 (2007) 98–105